SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2301.05231 · NeurIPS · 2023

Equivariant Representation Learning in the Presence of Stabilizers

Mike Holenderski, Danica Kragic, Giovanni Marchetti, Armando Luis, Rey Pérez, Dmitri Jarnikov

arXiv · PDF · Open in the Atlas

Code that ran

8 of 10 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 1 violated it and 0 have no contract verdict; 7 failures of our instrument; 2 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2301.05231.json · badge: /api/ran/2301.05231.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
luis-armando-perez-rey/non-free canonical (classifier label)
classifier basis: classifier_2025
8 ran · 2 no recorded run
FunctionStatusWhere it lives
azimuth_elevation_to_rotation_matrix Ran
ran; our contract draft was wrong, not the code
luis-armando-perez-rey/non-free/ENR/models/rotation_layers.py
code served (permissive licence) · get_code("1504c7de08612861")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 3, dtype family float, shape (2, 3, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from math import pi

torch.manual_seed(0)

# Create small batch tensors for azimuth and elevation
batch_size = 2
azimuth = torch.tensor([0.0, 45.0])
elevation = torch.tensor([30.0, 60.0])

ARGS = {
    "azimuth": azimuth,
    "elevation": elevation,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, dtype family float, shape (2, 3, 3) — shape (2, 3, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
deg2rad Ran
ran, violated the contract we drafted
luis-armando-perez-rey/non-free/ENR/models/rotation_layers.py
code served (permissive licence) · get_code("91f4176cd43c3ce2")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 1, dtype family float, shape (4,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np
from math import pi

torch.manual_seed(0)

# The function deg2rad takes angles and converts them from degrees to radians
# It simply multiplies by pi/180.0
# The parameter 'angles' can be a scalar, list, numpy array, or torch tensor

# Test with a torch tensor
angles_tensor = torch.tensor([0.0, 90.0, 180.0, 270.0])

ARGS = {
    "angles": angles_tensor
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 1, dtype family float, shape (4,) — shape (4,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_affine_grid Ran
ran; our contract draft was wrong, not the code
luis-armando-perez-rey/non-free/ENR/models/rotation_layers.py
code served (permissive licence) · get_code("e4f25344bf41ab44")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 5, dtype family float, shape (2, 4, 5, 6, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn.functional as F

torch.manual_seed(0)

# Construct the matrix argument: batch of 3x3 matrices
# batch_size=2, so shape should be (2, 3, 3)
matrix = torch.randn(2, 3, 3)

# Construct the grid_shape argument
# Should be of the form (batch_size, channels, depth, height, width)
# Using batch_size=2, channels=3, depth=4, height=5, width=6
grid_shape = torch.Size([2, 3, 4, 5, 6])

ARGS = {
    "matrix": matrix,
    "grid_shape": grid_shape,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 5, dtype family float, shape (2, 4, 5, 6, 3) — shape (2, 4, 5, 6, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
rotate Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
luis-armando-perez-rey/non-free/ENR/models/rotation_layers.py
code served (permissive licence) · get_code("9eaed97a3cf5dc5a")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 5, dtype family float, shape (2, 3, 4, 5, 6) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn.functional as F

torch.manual_seed(0)

# Create a small batch of 5D volumes: (batch_size, channels, depth, height, width)
# Using distinct sizes: batch=2, channels=3, depth=4, height=5, width=6
volume = torch.randn(2, 3, 4, 5, 6)

# Create rotation matrices: (batch_size, 3, 3)
# These should be orthogonal matrices, but for testing we'll use random ones
rotation_matrix = torch.randn(2, 3, 3)

# The mode parameter for grid_sample
mode = 'bilinear'

ARGS = {
    'volume': volume,
    'rotation_matrix': rotation_matrix,
    'mode': mode
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 5, dtype family float, shape (2, 3, 4, 5, 6) — shape (2, 3, 4, 5, 6) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
rotation_matrix_source_to_target Ran
ran; our contract draft was wrong, not the code
luis-armando-perez-rey/non-free/ENR/models/rotation_layers.py
code served (permissive licence) · get_code("66e42ff0a781b603")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 3, dtype family float, shape (2, 3, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from math import pi

torch.manual_seed(0)

# Create batch_size=2 tensors with azimuth and elevation angles in degrees
azimuth_source = torch.tensor([0.0, 45.0])
elevation_source = torch.tensor([0.0, 30.0])
azimuth_target = torch.tensor([90.0, 135.0])
elevation_target = torch.tensor([15.0, 60.0])

ARGS = {
    "azimuth_source": azimuth_source,
    "elevation_source": elevation_source,
    "azimuth_target": azimuth_target,
    "elevation_target": elevation_target,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, dtype family float, shape (2, 3, 3) — shape (2, 3, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
rotation_matrix_y Ran
ran; our contract draft was wrong, not the code
luis-armando-perez-rey/non-free/ENR/models/rotation_layers.py
code served (permissive licence) · get_code("15376cf46fc90fa6")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 3, dtype family float, shape (4, 3, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from math import pi

torch.manual_seed(0)

# The function rotation_matrix_y takes an angle tensor of shape (batch_size,)
# containing rotation angles in degrees
angle = torch.tensor([0.0, 45.0, 90.0, 180.0])

ARGS = {
    "angle": angle
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, dtype family float, shape (4, 3, 3) — shape (4, 3, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
rotation_matrix_z Ran
ran; our contract draft was wrong, not the code
luis-armando-perez-rey/non-free/ENR/models/rotation_layers.py
code served (permissive licence) · get_code("ef2076cd22670e8e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 3, dtype family float, shape (2, 3, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from math import pi

torch.manual_seed(0)

# The function rotation_matrix_z takes:
# - angle: torch.Tensor with shape (batch_size,) in degrees
# - device: optional device parameter (not used in function body)

# Create a small batch of angles in degrees
angle = torch.tensor([0.0, 90.0], dtype=torch.float32)

ARGS = {
    "angle": angle,
    "device": None
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, dtype family float, shape (2, 3, 3) — shape (2, 3, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
transpose_matrix Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
luis-armando-perez-rey/non-free/ENR/models/rotation_layers.py
code served (permissive licence) · get_code("6392f203079978bd")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 3, dtype family float, shape (2, 3, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Create a batch of matrices with shape (batch_size, n, m)
# Using distinct dimensions: batch_size=2, n=4, m=3
matrix = torch.randn(2, 4, 3)

ARGS = {
    "matrix": matrix
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, dtype family float, shape (2, 3, 4) — shape (2, 3, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
Rotate3d Not yet run
unverified (harvested, no recorded run)
luis-armando-perez-rey/non-free/ENR/models/rotation_layers.py
code served (permissive licence) · get_code("2532e3299f51d717")
How it ran

no invocation recorded.

rotate_source_to_target Not yet run
unverified (harvested, no recorded run)
luis-armando-perez-rey/non-free/ENR/models/rotation_layers.py
code served (permissive licence) · get_code("4d28e43f2eaa3ba7")
How it ran

no invocation recorded.

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 10 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

We introduce Equivariant Isomorphic Networks (EquIN)a method for learning representations that are equivariant with respect to general group actions over data. Differently from existing equivariant representation learners, EquIN is suitable for group actions that are not free, i.e., that stabilize data via nontrivial symmetries. EquIN is theoretically grounded in the orbit-stabilizer theorem from group theory. This guarantees that an ideal learner infers isomorphic representations while trained on equivariance alone and thus fully extracts the geometric structure of data. We provide an empirical investigation on image datasets with rotational symmetries and show that taking stabilizers into account improves the quality of the representations.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2301.05231")
get_code_for_paper("2301.05231")
have("2301.05231")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2301.05231.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2301.05231.svg)](https://syntology.ai/paper/2301.05231)

Connect an agent — have() is free.